product-video-creator

Multi-Agent Collaborative System for Product Video Creation

Task Objective

• This Skill is used to: transform product information and requirements into a complete product promotional video
• Capabilities include:
  • Copywriting: analyze product selling points and generate engaging core copy
  • Story planning: design scene storylines that fit the product
  • Scriptwriting: transform the story into an executable storyboard script
  • Storyboard design: plan the visual description, duration, and transitions for each shot
  • Image generation: generate scene images that match the description for the storyboard
  • Subtitle creation: create concise subtitle text for each shot
  • Sound effect recommendation: recommend background music and sound effect plans
  • Video composition: combine all materials into the final MP4 video
• Trigger conditions: the user provides product information (name, selling points, images, etc.) and video requirements (duration, style, target audience)

Preliminary Preparation

• Dependency description: dependencies required by the video composition script

  moviepy>=1.0.3
  pillow>=10.0.0
  opencv-python>=4.8.0
  

• File preparation: ensure the following resources are available in the working directory (if customization is needed)
  • bgm/: background music folder (optional; the script will recommend options, and defaults can also be used)
  • product_images/: original product images (optional; the agent can generate based on the description)

Operation Steps

Step 1: Copywriter - Core Selling Point Refinement

Responsibility: analyze product information and generate core copy that appeals to the target audience

Execution Points：

• Identify the product's core selling points and differentiated advantages
• Adjust the language style for the target audience (such as young women, family users, business professionals, etc.)
• Generate 3-5 alternative copy options, including:
  • Title (short and powerful, 5-10 characters)
  • Core slogan (catchy, 15-20 characters)
  • Detailed selling points (3-5 key points)

Output: core copy text

Step 2: Story Planner - Scene Story Design

Responsibility: design a coherent scene storyline based on the copy so that the product's selling points are naturally integrated

Execution Points：

• Design the story theme (such as "problem-solution", "usage scenario demonstration", "comparison test", etc.)
• Plan the story's introduction, development, turn, and conclusion, usually 3-5 scenes
• Ensure the story logic is smooth and the selling points appear naturally rather than being forcibly inserted
• Design the emotional curve (attention-interest-desire-action)

Output: scene story outline (including the plot and selling-point connection for each scene)

Step 3: Scriptwriter - Storyboard Script Writing

Responsibility: transform the story into a specific storyboard script

Execution Points：

• Allocate the time for each shot according to the video duration requirements (usually 8-12 shots for a 30-second video)
• Define for each shot:
  • Number (such as Shot 1, Shot 2)
  • Duration (seconds)
  • Description of the visual content (detailed down to characters, actions, background, product placement)
  • Subtitle content (short and concise, coordinated with the visuals)
  • Transition method (cut, fade in/out, push/pull, etc.)
• Ensure the storyboard continuity and sense of rhythm

Output: Storyboard script table, format reference references/storyboard_format.md

Step 4: Storyboard Director - Visual Design Optimization

Responsibilities: Generate a detailed visual description for each storyboard shot to guide image generation

Execution Points：

• Based on the storyboard script, create detailed visual description prompts for each shot
• The descriptions include:
  • Subject (product, character, environment)
  • Style (realistic, illustration, 3D rendering, etc.)
  • Lighting (natural light, indoor light, dramatic lighting)
  • Composition (close-up, medium shot, wide shot, perspective)
  • Color tone (warm, cool, high contrast, etc.)
• Ensure the visual style is consistent and matches the overall style of the video

Output: A list of visual description prompts for each shot

Step 5: Storyboard Artist - Scene Image Generation

Responsibilities: Generate images that match the descriptions for each storyboard shot

Execution Points：

• Use image generation capabilities to generate images based on the Storyboard Director's description prompts
• Ensure image quality: resolution of at least 1920x1080
• Maintain style consistency: all images use the same style parameters
• Highlight product display: ensure the product is clearly visible and reasonably positioned in the frame
• Generate multiple versions to choose from when necessary

Output: Image sequence, named scene_001.jpg, scene_002.jpg, ...

Step 6: Subtitle Specialist - Subtitle Creation and Design

Responsibilities: Create subtitle text for each shot and plan the display position and timing

Execution Points：

• Based on the subtitle content of the storyboard script, create the final subtitle text
• Ensure subtitles are short and easy to read (no more than 20 characters per line)
• Design the subtitle position (usually at the lower 1/3 of the frame)
• Plan the subtitle display timing (appear in the first 1/3 of the shot, disappear in the last 1/3)
• Consider subtitle color and border to ensure readability

Output: Subtitle text and timing configuration, format reference references/subtitle_format.md

Step 7: Sound Designer - Audio Solution Recommendation

Responsibilities: Recommend background music and sound effect solutions to enhance the video's appeal

Execution Points：

• Recommend background music types (upbeat, warm, dynamic, premium, etc.)
• Explain the emotional tone of the music and how it is used (used throughout, switched by segment, etc.)
• Recommend key sound effects (such as transition sound effects, click sounds, product entrance sounds, etc.)
• Provide music source suggestions (free asset libraries, royalty music platforms, etc.)
• Explain volume balance (background music volume should be lower than vocals/sound effects)

Output: Audio solution documentation

Step 8: Video Engineer - Video Synthesis

Responsibilities: Call scripts to synthesize images, subtitles, and audio into the final video

Execution Points：

• Ensure all materials are ready:
  • Image sequence:scene_*.jpg
  • Subtitle configuration:subtitles.json
  • Audio file:bgm.mp3(or other formats)
• Call the compositing script to handle technical details
• Verify video quality: resolution, frame rate, audio-video sync, subtitle clarity

Call script：

python scripts/video_composer.py \
  --images ./scene_*.jpg \
  --subtitles ./subtitles.json \
  --audio ./bgm.mp3 \
  --output ./final_video.mp4 \
  --fps 24 \
  --duration 30

Output: final video file final_video.mp4

Resource index

• Required script: see scripts/video_composer.py(Purpose: video material compositing and export)
• Storyboard script format: see references/storyboard_format.md(When to read: when writing the storyboard script in step 3)
• Subtitle format specification: see references/subtitle_format.md(When to read: when configuring subtitles in step 6)
• Video parameter specification: see references/video_specs.md(When to read: before compositing the video in step 8)

Notes

• Maintain collaboration between agents: the output of each step should become the input of the next step to ensure continuity
• Prioritize using agent capabilities: fully utilize the agent's existing capabilities for creative content, copywriting, image generation, and other stages
• Scripts focus on technical processing: only video compositing uses scripts; other stages are completed by agents
• Quality control: after each step is completed, the agent should self-assess the output quality and adjust if necessary
• Flexible adjustment: according to video duration and complexity, the number of shots can be increased or decreased, but logical completeness should be maintained

Usage examples

Example 1: 30-second new product promotion video

• Function description: create a 30-second promotional video for a new smartwatch
• Product information: smartwatch, supports health monitoring, ultra-long battery life, fashionable appearance
• Target audience: urban white-collar workers aged 20-35
• Video style: modern and minimalist, tech feel
• Execution process:
  1. Copywriter generates the core copy: "Smart life, controlled on your wrist"
  2. Story planner designs the "a day in life" scene
  3. Scriptwriter writes 8 storyboard shots (3-4 seconds each)
  4. Storyboard director designs visual descriptions in a modern tech style
  5. Storyboard artist generates 8 scene images
  6. Subtitle artist creates subtitles that match the visuals
  7. Sound designer recommends upbeat tech-style background music
  8. Video engineer calls the script to composite the video

Example 2: 60-second skincare product promotional video

• Function description: create a 60-second brand video for anti-aging skincare products
• Product information: high-end anti-aging essence, focusing on firming and brightening
• Target audience: women aged 30-50
• Video style: warm and elegant, premium feel
• Execution process:
  1. Copywriter: "Time passes, beauty does not age"
  2. Story planner designs a "before and after use comparison" story
  3. Scriptwriter writes 12 storyboard shots
  4. Storyboard director designs visuals with warm lighting and elegant style
  5. Storyboard artist generates 12 high-quality images
  6. Subtitle artist creates emotional subtitles
  7. Sound designer recommends soothing and elegant background music
  8. Video engineer composites the 60-second video

Example 3: 15-second social media short video

• Function description: create a 15-second Douyin-style short video for new snacks
• Product information: internet-famous snacks, rich flavors, exquisite packaging
• Target audience: young people aged 18-30
• Video style: lively and fun, fast-paced
• Execution process:
  1. Copywriter: concise and powerful slogan
  2. Story planner: quick presentation of a single scene
  3. Scriptwriter: 4-5 fast-paced storyboards
  4. Storyboard director: bright colors, dynamic composition
  5. Storyboard artist: generate 4-5 vivid images
  6. Subtitle artist: short subtitles matching the rhythm
  7. Sound effects artist: dynamic music paired with transition sound effects
  8. Video engineer: composite a 15-second video